The Consensus Coding Sequences of Human Breast and Colorectal Cancers Tobias Sjöblom,1* Siân Jones,1* Laura D
Total Page：16
File Type：pdf, Size：1020Kb
The Consensus Coding Sequences of Human Breast and Colorectal Cancers Tobias Sjöblom,1* Siân Jones,1* Laura D. Wood,1* D. Williams Parsons,1* Jimmy Lin,1 Thomas Barber,1 Diana Mandelker,1 Rebecca J. Leary,1 Janine Ptak,1 Natalie Silliman,1 Steve Szabo,1 Phillip Buckhaults,2 Christopher Farrell,2 Paul Meeh,2 Sanford D. Markowitz,3 Joseph Willis,4 Dawn Dawson,4 James K. V. Willson,5 Adi F. Gazdar,6 James Hartigan,7 Leo Wu,8 Changsheng Liu,8 Giovanni Parmigiani,9 Ben Ho Park,10 Kurtis E. Bachman,11 Nickolas Papadopoulos,1 Bert Vogelstein,1† Kenneth W. Kinzler,1† Victor E. Velculescu1† 1Ludwig Center and Howard Hughes Medical Institute, Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins, Baltimore, MD 21231, USA. 2Department of Pathology and Microbiology, Center for Colon Cancer Research, and South Carolina Cancer Center, Division of Basic Research, University of South Carolina School of Medicine, Columbia, SC 29229, USA. 3Department of Medicine, Ireland Cancer Center, and Howard Hughes Medical Institute, Case Western Reserve University and University Hospitals of Cleveland, Cleveland, OH 44106, USA. 4Department of Pathology and Ireland Cancer Center, Case Western Reserve University and University Hospitals of Cleveland, Cleveland, OH 44106, USA. 5Harold C. Simmons Comprehensive Cancer Center, University of Texas Southwestern Medical Center, Dallas, TX 75390, USA. 6Hamon Center for Therapeutic Oncology Research and Department of Pathology, University of Texas Southwestern Medical Center, Dallas, TX 75390, USA. 7Agencourt Bioscience Corporation, Beverly, MA 01915, USA. 8SoftGenetics LLC, State College, PA 16803, USA. 9Departments of Oncology, Biostatistics, and Pathology, Johns Hopkins Medical Institutions, Baltimore, MD 21205, USA. 10Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins, Baltimore, MD 21231, USA. 11Department of Radiation Oncology and Department of Biochemistry and Molecular Biology, Marlene and Stewart Greenebaum Cancer Center, University of Maryland School of Medicine, Baltimore, MD 21201, USA. *These authors contributed equally to this work. †To whom correspondence should be addressed. E-mail: [email protected]; [email protected]; [email protected] The elucidation of the human genome sequence has made diagnostic and therapeutic purposes. To date, however, only a it possible to identify genetic alterations in cancers in small fraction of the genes has been analyzed and the number unprecedented detail. To begin a systematic analysis of and type of alterations responsible for the development of such alterations, we have determined the sequence of well- common tumor types are unknown (2). In the past, the annotated human protein coding genes in two common selection of genes chosen for mutational analyses in cancer tumor types. Analysis of 13,023 genes in 11 breast and 11 has been guided by information from linkage studies in colorectal cancers revealed that individual tumors cancer-prone families, identification of chromosomal accumulate an average of ~90 mutant genes but that only abnormalities in tumors, or known functional attributes of a subset of these contribute to the neoplastic process. individual genes or gene families (2–4). The determination of Using stringent criteria to delineate this subset, we the human genome sequence coupled with improvements in identified 189 genes (average of 11 per tumor) that were sequencing and bioinformatic approaches have now made it mutated at significant frequency. The vast majority of possible, in principle, to examine the cancer cell genome in a these genes were not known to be genetically altered in comprehensive and unbiased manner. Such an approach not tumors and are predicted to affect a wide range of cellular only provides the means to discover other genes that functions, including transcription, adhesion, and invasion. contribute to tumorigenesis but can also lead to mechanistic These data define the genetic landscape of two human insights that are only evident through a systems biological cancer types, provide new targets for diagnostic and perspective. Comprehensive genetic analyses of human therapeutic intervention, and open fertile avenues for cancers could lead to discovery of a set of genes, linked basic research in tumor biology. together through a shared phenotype, that point to the importance of specific cellular processes or pathways. It is widely accepted that human cancer is a genetic disease To begin the systematic study of the cancer genome, we caused by sequential accumulation of mutations in oncogenes have examined a major fraction of human genes in two and tumor suppressor genes (1). These tumor-specific (that is, common tumor types, breast and colorectal cancers. These somatic) mutations provide clues to the cellular processes cancers were chosen for study because of their substantial underlying tumorigenesis and have proven useful for clinical significance world-wide: together, they account for / www.sciencexpress.org / 7 September 2006 / Page 1 / 10.1126/science.1133427 ~2.2 million cancer diagnoses (20% of the total) and considered successfully analyzed, we required that ≥90% of ~940,000 cancer deaths each year (14% of the total) (5). For bases in the target region have a Phred quality score (defined genetic evaluation of these tumors, we focused on a set of as –10[log10(raw per-base error)]) of at least 20 in at least protein coding genes, termed the consensus coding sequences three quarters of the tumor samples analyzed (8). This quality (CCDS) that represent the most highly curated gene set cutoff was chosen to provide high sensitivity for mutation currently available (6). The CCDS database contains full- detection while minimizing false positives. Using these length protein coding genes that have been defined by criteria, 93% of the 135,483 amplicons and 91% of the total extensive manual curation and computational processing and targeted bases in CCDS were successfully analyzed for have gene annotations that are identical among reference potential alterations. databases. Examination of sequence traces from these amplicons The goals of this study were three-fold: (i) to develop a revealed a total of 816,986 putative nucleotide changes. As methodological strategy for conducting genome-wide the vast majority of changes that did not affect the amino acid analyses of cancer genes in human tumors; (ii) to determine sequence (i.e., synonymous or silent substitutions) were likely the spectrum and extent of somatic mutations in human to be non-functional, these changes were not analyzed further. tumors of similar and different histologic types; and (iii) to The remaining 557,029 changes could represent germline identify new cancer genes and molecular pathways that could variants, artifacts of PCR or sequencing, or bona fide somatic lead to improvements in diagnosis or therapy. mutations. Several bioinformatic and experimental steps were Cancer mutation discovery screen. The initial step employed to distinguish among these possibilities. First, any toward achieving these goals was the development of alterations that were also present in either of the two normal methods for high-throughput identification of somatic samples included in the Discovery Screen were removed, as mutations in cancers. These methods included those for these were likely to represent common germline primer design, polymerase chain reaction (PCR), sequencing, polymorphisms or sequence artifacts. Second, as these two and mutational analysis (Fig. 1). The first component normal control samples would be expected to contain only a involved extraction of all protein coding sequences from the subset of known variants, any change corresponding to a CCDS genes. A total of 120,839 non-redundant exons and validated germline polymorphism found in single nucleotide adjacent intronic sequences were obtained from 14,661 polymorphism (SNP) databases was also removed (7). different transcripts in CCDS. These sequences were used to Finally, the sequence trace of each potential alteration was design primers for PCR amplification and sequencing of visually inspected in order to remove false positive calls in exons and adjacent splice sites. Primers were designed using the automated analysis. The combination of these data a number of criteria to ensure robust amplification and analysis efforts was efficient, removing ~96% of the potential sequencing of template regions (7). While most exons could alterations and leaving 29,281 for further scrutiny (Fig. 1). be amplified in a single PCR reaction, we found that exons To ensure that the observed mutations did not arise larger than 350 bp were more effectively amplified as artifactually during the PCR or sequencing steps, the regions multiple overlapping amplicons. One member of every pair of containing them were independently re-amplified and re- PCR primers was tailed with a universal primer sequence for sequenced in the corresponding tumors. This step removed subsequent sequencing reactions. A total of 135,483 primer 9,295 alterations. The regions containing the putative pairs encompassing ~21 Mb of genomic sequence were mutations were then sequenced in matched normal DNA designed in this manner (table S1). samples to determine whether the mutations were truly Eleven cell lines or xenografts of each tumor type (breast somatic: 18,414 changes were observed to be present in the and colorectal carcinomas) were used in the Discovery Screen germline of these patients, representing variants not currently (table S2, A and B). Two matching